|
Dettagli:
|
| FP64: | 34 teraFLOPS | Il centro del tensore FP64: | 67 teraFLOPS |
|---|---|---|---|
| FP32: | 67 teraFLOPS | Il centro del tensore TF32: | 989 teraFLOPS |
| Memoria di GPU: | 80GB | Larghezza di banda di memoria di GPU: | 3.35TB/s |
| Decodificatori: | 7 NVDEC 7 JPEG | Potere termico massimo di progettazione (TDP): | Fino a 700 W (configurabile) |
| GPU multiistanza: | Fino a 7 MIG da 10 GB ciascuno | Fattore di forma: | SXM |
| Evidenziare: | 80GB GPU Memory NVIDIA H100 SXM GPU,SXM Form Factor H100 SXM Tensor Core GPU,700W Configurable TDP NVIDIA H100 SXM GPU |
||
H100 features fourth-generation Tensor Cores and a Transformer Engine with FP8 precision that provides up to 4X faster training over the prior generation for GPT-3 (175B) models. For high-performance computing (HPC) applications, H100 triples the floating-point operations per second (FLOPS) of double-precision Tensor Cores, delivering 60 teraflops of FP64 computing for HPC while also featuring dynamic programming (DPX) instructions to deliver up to 7X higher performance. With
Persona di contatto: Alisa CC
Telefono: +86 18710168043